namb physics parameterizations - richard...
TRANSCRIPT
NAMB Physics Parameterizations
COMET & Jimy Dudhia (modified by Richard Grotjahn)
Winter, 2017
Sources: • Borrowed heavily from
this NCEP document and the meted Operational Models Encyclopedia.
• http://www.mmm.ucar.edu/wrf/users/tutorial/200807/WRF_Physics_Dudhia.pdf
• https://sites.google.com/a/ucar.edu/model-encyclo-determ/deterministic/namb
Overview • A ‘physics parameterization’ is an approximation of a
process. • They are often not of direct interest to a forecaster.
(primary exception: precipitation) • Process parameterized to obtain with ‘sufficient
accuracy’ the impact of that process upon a primary variable. (e.g. how radiation affects primary variable T)
• Various approximations (schemes) exist for the same physical process. Different schemes have different advantages/disadvantages: speed, accuracy in certain situations, storage space, etc.
• Generally speaking, the physics parameterizations are more accurate in the regional model than the global model. (partly due to better resolution)
• You do not need to memorize the model details that follow. Just learn the types of processes and issues.
Outline • Overview • Microphysics in clouds (heat & moisture
tendencies; rates; other surface precipitation) • Convective parameterization (heat & moisture
tendencies; surface precipitation) • Radiation (longwave & shortwave; clear) • Radiation (cloud effects) • Surface schemes (fluxes, land surface models:
LSMs, SST) • Turbulence & diffusion • Planetary boundary layer (PBL fluxes, vertical
diffusion) • Interactions between physics parameterizations • Summary
Precipitation & Clouds-1 • The most important features of the scheme are:
– It allows hydrometeors to grow and shrink in size – It uses an ice density to account for different
forms of frozen precipitation – Fall speeds depend on hydrometeor size and type
• Sedimentation/sorting occurs as larger and especially denser particles fall faster
• Collection growth occurs as hydrometeors sweep out others on the way down
• Precipitation gradually falls to the ground at appropriate speeds and can be advected as part of the total condensate while falling
– It allows mixed phase precipitation, including • Riming • Shedding of rain off melting ice • Coexistence and interaction of all forms of liquid and
frozen cloud and precipitation particles under certain conditions
Precipitation & Clouds -2 • These features allow the model to predict a variety of
phenomena: – Snow blowing over a ridge crest or downstream from the
Great Lakes – Density of frozen hydrometeors, which might, after further
development and then testing, be able to be used to predict snow:water ratios
– Good physical realism of precipitation type (caveat: consistent with model's temperature profile, which may differ from the real-life temperature profile!)
– Feeder-seeder precipitation mechanism – Cloud cover at different levels and different types of clouds
(water, ice, mixed) – More realistic cloud water content than previous scheme – Cirrus which precipitates out (fall streaks) – Virga and evolution from virga to precipitation on the ground – Warm rain processes (no ice) and cold rain processes (start
with snow)
Precipitation & Clouds -3 • The scheme assumptions do not allow prediction of
some phenomena: – Heavier precipitation associated with dendritic growth – Hail (riming included but no liquid below -40°C, limited
particle size and fall velocity) – Growth of snow falling through a temperature inversion
(snow will shrink slightly instead) – Effects of different aerosol types and concentrations,
including marine-continental differences in drop size distribution and precipitation production
– Convective cloud microphysics, featuring water supersaturation in updraft cores and unsaturated downdrafts side by side, easily fitting within a model grid column
– Separate advection of different hydrometeor types (mainly affects very high resolution runs)
Precipitation & Clouds -4 • an ensemble of FIVE precipitation type algorithms
are used and the predominant precipitation type is output to the forecast file. A graphic showing these precipitation types and how they work is shown below.
Precipitation & Clouds -5 • Process of the ‘PCP’
(precipitation and cloud parameterization) scheme
• Model dynamics and physics: model's wind, temperature, and water vapor mixing ratio fields. Any existing clouds and falling precip are advected.
• Instability relieved by CP (convective parameterization): heat and water vapor redistributed in the grid column, affecting the RH in model layers between LCL and equilibrium level. No condensate is generated; all convective precipitation falls instantly.
Precipitation & Clouds -6 • If RH(sat)ice < RH <
RH(sat)water: RH greater than 100% with respect to ice but less than 100% with respect to water, cloud ice and precipitation ice are added to any already present.
• If RH > RH(sat)water and T > −40°C: If RH exceeds 100% with respect to water within a three-dimensional grid box, cloud liquid water is added to any already present. Some cloud ice will form, with greater amounts closer to −40°C.
• If RH < threshold RH: If RH less than RH(sat)ice, no new cloud liquid and/or cloud ice are permitted but falling precipitation generated earlier may be present.
Precipitation & Clouds -7
• Precipitation generated earlier continues falling toward ground: Precipitation falls at realistic speeds. Light precipitation may take an hour or more to reach ground while heavy rain may reach the ground in a few minutes. Also, some precipitation evaporates.
• Cloud ice and frozen precipitation form: Cloud ice and snow form together, ranging from almost equal parts cloud ice and tiny snow crystals for cold cirrus to mainly large snowflakes at warmer temperatures.
Precipitation & Clouds -8
• Microphysical interaction of all condensate forms: Precip falling from above, pre-existing cloud ice and cloud water, and newly formed condensate all interact, including forming additional precip by collecting cloud water and cloud ice.
• Precipitation falls at terminal velocity: Light precip may take 1+ hour to reach ground while heavy rain may reach ground in a few minutes.
• Autoconversion starts immediately: Rates of autoconversion slower for ice at temperatures near 0°C and for low amounts of cloud liquid or ice, with a rapid, nonlinear increase with cloud water increase..
• Dynamics and physics continue: Rates of cloud condensation and precip production remain unchanged for 5-9 dynamical time steps (varies from NAMB parent and among inner nests), then get updated again using new RH and hydrometeor fields.
Precipitation & Clouds -9
• Summary.
Precipitation & Clouds -10
• Summary.
Convective Parameterization • Superadiabatic profiles of temperature can
form from advection, heating below, cooling above, etc.
• Various situations trigger convection • NAMB uses Betts-Miller-Janjic (‘BMJ’)
scheme – adjusts the model temperature and moisture
profiles toward reference profiles with temperature resembling a moist adiabat and dewpoint depressions of 4 to 6°C
Convection • Model dynamics and physics: Change winds,
temperature, and the water vapor mixing ratio fields. Any existing clouds are advected.
• If deep convection is triggered: Reference temperature and moisture profiles are determined and the model sounding is adjusted slightly toward the reference profiles, such that if the adjustments continued over a convective timescale of ~40 to 50 minutes, it would reach the reference profiles.
• Convective precipitation is generated and fall instantly to the ground: No microphysics are involved, no evaporation of convective precipitation occurs, and no convective cloud water is created. Convective clouds are flagged for using the radiation scheme.
• PCP scheme is run: The cloud and precipitation scheme runs using the adjustments to the RH field (if any) created by the convective scheme.
• If deep convection is not triggered: Shallow convection may be triggered for mixing through the top of the boundary layer (for instance, like the effects of moderate cumulus).
• Shallow convection is checked: If activated, the shallow CP scheme modifies the top of the boundary layer acting as an extension of the model physics boundary-layer mixing process.
• Dynamics and physics continue: Convective precipitation and convective sounding changes continue at the same rate for a total of nine minutes (six dynamic time steps), then get updated again using new model soundings.
Convective Tuning
parameters
• Values that are set to improve the forecast
• ‘Current’ (2/2017) values shown
• Other parameters are also available for change
Convective adjustment -1 • Part of the model sounding (yellow dashed, ‘Y-shaped’) forced
towards reference profiles (blue). • Rate of adjustment varies with intensity of the precip, location,
etc.
Convective adjustment -2 • Reference profiles constructed from first guess T, Td that are adjusted • For T: criteria applied in different T ranges (e.g. T<0C) • Based on moisture change, T remains warmed as in lifted parcel (middle, blue) or
must be cooled (purple) to have ‘same’ amounts of cooling and heating layers • For Td: this adjustment determines amount of precip generated (which may have
different types, interact, etc.) Deficit saturation pressure (DSP) at 3 levels (right fig) which specify the shift of the profile based on the adjusted T profile
Radiation
• Clear sky • Cloudy sky
Solar Radiation – clear sky • Absorbers:
– Seasonal prescribed ozone – Predicted water vapor – Prescribed CO2 – Oxygen (diatomic) – Aerosols
• Shortwave spectrum divided into halves of energy (Lacis and Hansen scheme)
• Amount of absorber estimated in each layer, then absorption accumulates as sunbeam heads down
• Reflection & scattering diminish the beam
• Varies diurnally and seasonally • Is a heating term in the
temperature tendency eqn. • Impacts the surface energy budget
(radiation reaching sfc)
Solar Radiation – cloud • Assumptions about opacity differ for different cloud types
(see fig) • Reflects, absorbs, transmits sunbeam • Hydrometeors included • Again, T tendency in layers and sfc energy budge impacted
Shortwave radiation (NMM)
Terrestrial Radiation – clear sky • Longwave radiation, LWR • Absorbers:
– Seasonal prescribed ozone – Predicted water vapor – Prescribed CO2
• Model sfc radiates as black body (calc. every time step)
• Absorption in bands (fig) Schwarzkopf and Fels scheme. Bands overlap for different gases.
• Layers absorb & emit both upwards and downwards
• Impacts temperature tendency in layers
• Impacts surface energy budget
Terrestrial Radiation – cloud • Assumptions about optical depth differ for different cloud
types (see fig) • Absorbs, emits, transmits longwave radiation (up & down) • Hydrometeors included • Again, T tendency in layers and sfc energy budge impacted
Longwave radiation (NMM)
Interval between radiation calculations
Surface models (land, water, snow)
• Land models – Vegetation type – Soil moisture
• Water – Lakes – Ocean – Sea ice
• Snow cover
land • Soil layers
– T, moisture, etc. • Vegetation type
– Transpiration, albedo, etc.
water • Water
– Lakes – ocean
• Prescribed water surface T changes slowly
• daily values for large bodies: ocean, great lakes, Salton sea
• Surfaces specified (map) • Sea and lake ice
specified • Water has low albedo
except for low sun angles
Snow cover, ice cover • Albedo • Fractional (patchy) coverage • Snowpack evolution (density,
thermal conductivity, depth) function of time, past conditions, vegetation type, ...
• Sea ice (daily) analysis, but fixed during integration (affects albedo, sfc budget, etc.)
Turbulence & PBL
• Turbulence • Surface
energy budget
• Planetary boundary layer (PBL)
Turbulence & Topography • Turbulent diffusion (moisture, heat, momentum)
function of wind shear and static stability • Topographically generated
– Gravity wave drag – Flow blocking
Surface Energy Balance • Thermal balance in ‘microlayer’ (not skin
temperature) • SWR + LWR down = LWR up + LH + SH + GH • Figs have some details, e.g. no wind in microlayer. • Constraints on max LH, emissivity, plant ET, …
Planetary Boundary Layer • Diffusion through the PBL uses
modified Mellor-Yamada 2.5-order scheme. The amount of TKE determines the amount of mixing, and in turn the TKE increases from its production and reduces through dissipation.
• TKE production results from – Vertical wind shear – Buoyancy/instability
• TKE dissipation results from – Eddy and molecular viscosity – Surface friction
• Production and dissipation parameterized. Model production of TKE is function of diffusion rate (momentum, heat, or moisture) and
– Differences in zonal and meridional wind speeds across a model layel
– Differences in potential temperature across a model layer
• The diffusion rates in each layer depend on the layer TKE and other empirically determined stability parameters
• TKE production from vertical gradients of wind and T (mechanical and buoyant production). Dissipation from molecular, eddy, and surface friction are calculated first. Resulting TKE values used to adjust the vertical diffusion rates and changes in TKE resulting from vertical diffusion.
Summary • A ‘physics parameterization’ is an approximation of a
process. • They are often not of direct interest to a forecaster.
(primary exception: precipitation) • Process parameterized to obtain with ‘sufficient
accuracy’ the impact of that process upon a primary variable. (e.g. how radiation affects primary variable T)
• Various approximations (schemes) exist for the same physical process. Different schemes have different advantages/disadvantages: speed, accuracy in certain situations, storage space, etc.
• Generally speaking, the physics parameterizations are more accurate in the regional model than the global model. (partly due to better resolution)
